Definition of Biostatistics
David Jonga
"Biostatistics comprises statistical methods that are used to manage uncertainties in the field of medicine and health." (Indrayan A. Medical Biostatistics, Second Edition, Chapman & Hall/CRC Press, 2008: p2). This book defines statistics as the sciences of management of uncertainties--a tool to measure them and minimize their impact on decisions. A tall order indeed but perhaps understandable.
What are the other one line definitions that adequately describe the subject matter of biostatistics?
Ted Harding
The quoted definition is grossly restrictive.
Biostatistics embraces all aspects of statitsics which are applicable
to any domain of biology.
There's your one-liner! Of course this includes medicine and health,
but also covers everything from the genetics of fruit-flies to the
migration of eels to the development of genetically modified food
crops to the sustainability of the whaling industry to the guilt
of suspects identified from DNA samples to the lethality of Sarin
nerve-agent to ...
As to statistics, there isn't really room here! "Management of
uncertainties" is not bad as a rather vague blanket term, but
focussing on decision (as the book's statement appears to do)
is too restrictive -- whether a suspect is guilty is undoubtedly
a matter for decision, but the question "What do we know about
the migration of eels?" is not.
A better-focussed description of what statistics is about could
be that it deals with information from source to sheet of paper.
"Information" includes the notion of uncertainty -- unpredictability
of what information may be emitted by a source under given
conditions, uncertainty about what corruption of the infromation
may occur along the line, uncertainty about what additional
irrelevant information ("noise") may be added along the line.
Statistics, in the broadest sense, is the science of extracting
the maximum amount of relevant information from available data,
and (where applicable) of designing the system for obtaining
data so as to maximise the relevant information content.
How this is achieved in any particular case depends on what
specific interpretations are given to the very general terms
in the above.
Clearly, it is a huge subject!
Ted.
--------------------------------------------------------------------
E-Mail: (Ted Harding) <Ted.H...@manchester.ac.uk>
Fax-to-email: +44 (0)870 094 0861
Date: 11-Oct-08 Time: 08:52:48
------------------------------ XFMail ------------------------------
BXC (Bendix Carstensen)
But one question still remains: Why "biostatistics". Why is the term of interest as opposed to for example statistics used in analysis of data on hardening of reinforced concrete?
My bet is that once in the "bio"-sphere one can only ever hope to model a small fraction what really goes on in any one observed system. Which implies that the between individual variation always will be a one of the major contributors to uncertainty. Hence a central part biostatistics is to identify this variance component and separate it from other sources of variation.
Bendix
______________________________________________
Bendix Carstensen
Senior Statistician
Steno Diabetes Center
Niels Steensens Vej 2-4
DK-2820 Gentofte
Denmark
+45 44 43 87 38 (direct)
+45 30 75 87 38 (mobile)
b...@steno.dk http://www.biostat.ku.dk/~bxc
> applicable to any domain of biology.x
Ted Harding
one can only ever hope to model a small fraction what really goes on"
One aspect of this is between-individual variation (why does one
patient die and another recover when you could not see how they
might differ?). And the "elimination" of such variation certainly
led to well-known techniques -- the effect of a drug on say
blood pressure is more accurately estimated by measuring BP on
a set of subjects, administering the drug, measuring BP afterwards,
and then using the difference within each subject as the data
("paired t-test" approach) rather than measuring BP on one set
of people who have not taked the drug, and measuring it on another
set who have taken the drug.
But there is a lot more to the "biological unobservability" issue
than this! A good (and classical) example is the estimation of the
size of (say) the population of Cod in the North Sea. You certainly
can't go down under the water with a clipboard and count them.
In early days, the "mark-recapture" technique was developed: Take the
ship out, catch some Cod (M in number), clip a tag onto each fish,
and return it to the sea. Come back later, catch some Cod (n in number)
and count how many (m) are tagged.
*IF* you assume that the M you first caught are still alive, still
have their tags, and have uniformly dispersed through the population,
then the occurrence of m tagged fish in your n 2nd-time catch,
out of your original M tagged fish, is equivalent to a random
sample of n out of N (the unkown population size) of which M are
marked (tagged) and (N-M) unmarked. This has a definite probability
which depends on m, M, n and N, and the only unkown is N. Hence you
have a definite channel of information about N (which is what you want
to know).
The following vivid passage from "Fish Migration" (F.R. Harden Jones,
1968) illustrates complications which can assault the above simple
approach:
"The mortality of tagged fish and the loss of tags
[...]
MORTALITY OF TAGGED FISH
The heaviest mortality is almost certainly due to the damage
done to the fish when they are caught [first caught, for tagging].
Most fish have a gas-filled swim-bladder, which functions as a
hydrostatic organ, making the density of the fish more or less
equal to that of the water in which it is living. When the fish
are brought to the surface from deep water the reduction in
hydrostatic pressure, 1 atmosphere for every 10m of water,
leads to an expansion of the swim-bladder gas. In some fish
the swim-bladder is open and has a duct opening into the
anterior portion of the gut. Excess gas can be released through
this duct which can be thought of as a sort of escape valve.
Herring, salmon and eels have open swim-bladders of this type.
In other fish, such as cod, haddock, and hake, the swim-bladder
is completely closed and the excess gas cannot get away unless
it is reabsorbed by the blood, which is a very slow process.
Cod taken from 40 to 80m often come up completely bloated,
and float helplessly on the water. They are in a very poor
condition for tagging. Surprisingly, cod caught in deeper
water, down to 200m or so, are in much better condition.
This is because the pressure of the expanded gas ruptures
the swim-bladder wall. The gas passes into the body cavity,
tearing the body wall near the anus, and escapes to the outside.
The wound, unless it is very big, does not seem to harm the cod,
but it may to some extent contribute to the mortality of those
that are tagged. Cod taken from water much deeper than 200m
are usually very badly blown and the swim-bladder gas does a
lot of damage before it escapes. There are severe internal
haemorrhages, the stomach may be forced out through the mouth,
the rectum everted.
[...]
Plaice, like other flatfish, do not have a swim-bladder,
and this sort of problem does not arise when tagging them."
And that's not the only thing that can go wrong ... A later passage:
"The interpretation of results
Tåning raised the interesting point that the recoveries of tagged
fish taken from deep water might be misleading if the fish did
not return to the the dpeths from which they were taken. Water
currents near the surface and the bottom usually move at different
speeds and sometimes in opposite directions. If the fish stayed
near the surface they might be carried miles away from the main
body whose migrations the experiments were designed to follow."
I know that Fish Migration is not Medicine, but I'm quoting the
above (which are among my favourite quotations) because Fisheries
scientists have a tradition of facing up to such problems and
discussing them in explicit detail. They therefore bring nicely
the surface (as it were ... ) some of the issues that arise when
you view (see my initial response below) statistics as the science
of what happens to information "from source to sheet of paper".
I wonder to what extent this happens in Medical Statistics?
Certainly some important publications (including ones from people
on this list) do take it seriously; but in the mass of routinely
reported investigations I do not often see such things mentioned.
The general assumption seems to be that "Statistics just works"
(rather like one's SatNav -- but SatNav horror stories would be
getting really off-topic, wouldn't they?).
Ted.
Date: 11-Oct-08 Time: 11:08:00
------------------------------ XFMail ------------------------------
Martin Holt
David Jonga
Martin Holt
----- Original Message -----From: David Jonga
Sent: Saturday, October 11, 2008 5:07 PMSubject: {MEDSTATS} Re: Definition of Biostatistics
BXC (Bendix Carstensen)
such lametations on the potential un-interpretability of the observed data are very common in medical statistics, particularlybin the field og physiology.
Bendix
David Jonga
John Whittington
I certainly don't understand what you mean by "Perhaps examples can be constructed that use statistical methods without using the concepts of probability and consequently not related to uncertainty." Can you give any examples of this kind of statistics ? (I've always thought that uncertainty lies at the root of statistics.)
Furthermore, in the wider world, I think that the most (if anything) that the general public know about the existence of a probability-based subject called 'Statistics' concerns the 'margins of error' sometimes quoted in relation to polls and surveys. Beyond that, they think of 'Statisticians' (hence Statistics) merely as those who present data. The media 'Statisticians' who often appear in relation to sporting activities etc. are probably closer to Historians (or Librarians, or databases!) than any sort of Statistician that you and I would recognise - since they merely 'spirit up' historical information.
Kind Regards,
Swank, Paul R
I have always been intrigued by the term biostatistics. On the surface, it seems to means statistics performed in the realm of the biomedical sciences but then we don’t have psycho-statisticians (for obvious reasons, or do we?), or socio-statisticians, or busi-statisticians (busy, maybe). Why does biostatistics have to be separated out? Of course there are the econometricians but whatever do they do? This is perhaps why I call myself a quantitative methodologist. My $.02 worth.
Paul
Paul R. Swank, Ph.D
Professor and Director of Research
Children's Learning Institute
University of Texas Health Science Center
Houston, TX 77038
From:
MedS...@googlegroups.com [mailto:MedS...@googlegroups.com] On Behalf Of John
Whittington
Sent: Monday, October 13, 2008 8:18 AM
To: MedS...@googlegroups.com
Subject: {MEDSTATS} Re: Definition of Biostatistics
At 19:04 11/10/2008 +0100, Martin Holt wrote (in part):
johnstep...@gmail.com
Sent from my BlackBerry® wireless handheld
From: "Swank, Paul R" <Paul.R...@uth.tmc.edu>
Date: Mon, 13 Oct 2008 15:11:19 -0500
Adrian Sayers
panel data, instrumental variables, etc. etc. etc.
Social statistics, just went to a conference the "RC33" lots of
graphical models and structural equation models, very focus on
evaluating fit, stuff i have never seen before as a medical
statistician. Interesting elements on probabilistic record linkage.
Mlwin was developed by social statisticians (goldstein et al.) to try
and analyze school data.
Demography- statistics relating to population, something to do with geography.
psychometricians- statistics in psychology??
medical statistics- statistics relating to medical issues.
bio statistics uses elements from all of these disciplines dont they.
Seems like a bit of an old generic term to me.
Adrian
2008/10/13 <johnstep...@gmail.com>:
> How can I insiscribe to this
>
> Sent from my BlackBerry(R) wireless handheld
Martin Holt
Swank, Paul R
Paul
Paul R. Swank, Ph.D
Professor and Director of Research
Children's Learning Institute
University of Texas Health Science Center
Houston, TX 77038
-----Original Message-----
From: MedS...@googlegroups.com [mailto:MedS...@googlegroups.com] On
Behalf Of Adrian Sayers
Sent: Tuesday, October 14, 2008 2:50 AM
To: MedS...@googlegroups.com
Subject: {MEDSTATS} Re: Definition of Biostatistics
Econometricians trickle down lots and lots of very interesting work on
panel data, instrumental variables, etc. etc. etc.
Social statistics, just went to a conference the "RC33" lots of
graphical models and structural equation models, very focus on
evaluating fit, stuff i have never seen before as a medical
statistician. Interesting elements on probabilistic record linkage.
Mlwin was developed by social statisticians (goldstein et al.) to try
and analyze school data.
"But they don't call them socio-statisticians!"
Demography- statistics relating to population, something to do with
geography. "But not demo-statisticians!"
psychometricians- statistics in psychology?? "Psychometricians are
experts in measurement theory, not statisticians."
medical statistics- statistics relating to medical issues. "but these
guys and gals are call 'Biostatisticians'!"
bio statistics uses elements from all of these disciplines dont they.
Seems like a bit of an old generic term to me.
Adrian
Seems to me that there are statisticians and biostatisticians. So why do
they get their own name? :>)
Adrian Sayers
and some that do both.
I am not a theoretical statistician, i just aint bright enough :(
But i am very happy trickling down the good stuff :)
Adrian
2008/10/14 Swank, Paul R <Paul.R...@uth.tmc.edu>: